The postnatal role of Sox9 in cartilage - Wiley Online Library

5 downloads 1021 Views 3MB Size Report
Jul 6, 2012 - different genotypes; namely, the control group Sox9flox/flox and ...... Haag J, Gebhard PM, Aigner T. SOX gene expression in human.
JBMR

ORIGINAL ARTICLE

The Postnatal Role of Sox9 in Cartilage Stephen P Henry , 1 Shoudan Liang , 2 Kadir C Akdemir , 3 and Benoit de Crombrugghe 1 1

Department of Genetics, The University of Texas M.D. Anderson Cancer Center; Houston, TX, USA Department of Biostatistics and Computational Biology, The University of Texas M.D. Anderson Cancer Center; Houston, TX, USA 3 Department of Biochemistry and Molecular Biology, The University of Texas M.D. Anderson Cancer Center; Houston, TX, USA 2

ABSTRACT Sox9 is an essential transcription factor for the differentiation of the chondrocytic lineage during embryonic development. To test whether Sox9 continues to play a critical role in cartilaginous tissues in the adult mice, we used an inducible, genetic strategy to disrupt the Sox9 gene postnatally in these tissues. The postnatal inactivation of Sox9 led to stunted growth characterized by decreased proliferation, increased cell death, and dedifferentiation of growth plate chondrocytes. Upon postnatal Sox9 inactivation in the articular cartilage, the sulfated proteoglycan and aggrecan content of the uncalcified cartilage were rapidly depleted and the degradation of aggrecan was accompanied by higher ADAMTS5 immunostaining and increased detection of the aggrecan neoepitope, NITEGE. In spite of the severe loss of Collagen 2a1 mRNA, the Collagen II protein persisted in the articular cartilage, and no histopathological signs of osteoarthritis were observed. The homeostasis of the intervertebral disk (IVD) was dramatically altered upon Sox9 depletion, resulting in disk compression and subsequent degeneration. Inactivation of Sox9 in the IVD markedly reduced the expression of several genes encoding extracellular matrix proteins, as well as some of the enzymes responsible for their posttranslational modification. Furthermore, the loss of Sox9 in the IVD decreased the expression of cytokines, cell-surface receptors, and ion channels, suggesting that Sox9 coordinates a large genetic program that is instrumental for the proper homeostasis of the cells contained in the IVD postnatally. Our results indicate that Sox9 has an essential role in the physiological control of cartilaginous tissues in adult mice. ß 2012 American Society for Bone and Mineral Research. KEY WORDS: Sox9; OSTEOARTHRITIS; INTERVERTEBRAL; AGGRECAN; COLLAGEN II

Introduction

A

major role of the transcription factor, SOX9, during chondrogenesis was strongly suggested by the identification of heterozygous mutations at the SOX9 gene locus in the human autosomal dominant disease, campomelic dysplasia (CD).(1,2) This disease, which is caused by haploinsufficiency of SOX9, is characterized by severe skeletal malformations such as shortening and bowing of the long bones, cervical spine flexure, and scoliosis. The requirement of Sox9 for cartilage formation was demonstrated by studies of mouse embryo chimeras in which Sox9/ cells were excluded from chondrogenic mesenchymal condensations but were found at surrounding, adjacent mesenchyme that did not express chondrocyte-specific genes.(3) Furthermore, the conditional homozygous ablation of the Sox9 gene during different stages of mouse embryonic limb development demonstrated that Sox9 is necessary for mesenchymal condensations in which the commitment to the chondrocyte fate is established from a common osteochondroprogenitor cell.(4) The manipulation of genetically engineered

mice has been a very valuable resource to delineate the role of Sox9 in cartilage formation during embryonic cartilage development. Understanding the role of Sox9 in postnatal and in adult cartilage tissues requires an inducible Cre recombinase. Note that the ability to inactivate genes in cartilage postnatally with an inducible genetic system is relevant because the articular cartilage along with the secondary ossification center is only formed postnatally. Previously described cartilage-specific inducible cre recombinase transgenes include the minimal Collagen 2a1 (Col2a1) promoter transgenic lines such as Col2a1-CreER,(5) Col 2a1-CreERT2,(6) Col2a1-rtTA,(7) and the targeted Agc1CreERT2 knockin allele.(8) The Sox9 mRNA in the articular joint cartilage of mice is detected at several time points of postnatal growth up to 9 months, with the highest Sox9 mRNA levels at the early phase of rapid growth.(9) Also, Sox9 protein was detected by immunohistochemistry in the mature articular cartilage at several time points including 15-month-old mice.(9) Gene expression analysis of normal and osteoarthritic adult human articular cartilage

Received in original form February 9, 2012; revised form June 5, 2012; accepted June 22, 2012. Published online July 6, 2012. Address correspondence to: Stephen P Henry, PhD, The University of Texas M.D. Anderson Cancer Center; Houston, TX 77030, USA. E-mail: [email protected] Additional Supporting Information may be found in the online version of this article. Journal of Bone and Mineral Research, Vol. 27, No. 12, December 2012, pp 2511–2525 DOI: 10.1002/jbmr.1696 ß 2012 American Society for Bone and Mineral Research

2511

revealed that Sox9 mRNA was reduced in the human osteoarthritic tissue.(10) Given the essential role of Sox9 in cartilage formation during embryonic development, one can postulate that Sox9 should play an important role in the homeostasis of postnatal and adult cartilage tissues. The transcription factor Sox9 is expressed in both the articular and growth plate hyaline cartilages as well as the intervertebral disk (IVD) of the spine.(11) The IVD consists of a dense connective tissue, which separates individual vertebrae along the spinal column. These IVDs serve as a cushion that permits mobility of the vertebral column and also as shock absorbers that protect the vertebral bones. Furthermore, it has been shown that SOX9 is required for notochord maintenance and normal vertebral column development during mouse embryonic development.(12) Also, the levels of Sox9 in the human IVD decrease upon disk degeneration.(11,13) To test the role of Sox9 in postnatal growth and cartilage homeostasis, we have bred mice containing the homozygous conditional Sox9 allele (Sox9flox/flox)(4) to the Aggrecan-CreERT2 knockin allele,(8) thus generating Agc1CreERT2; Sox9flox/flox mice. These mice were injected with the inducer, tamoxifen, at different points during postnatal development resulting in the conditional inactivation of the Sox9 gene in adult mice, or Agc1CreERT2; Sox9D/D mice. The disruption of the Sox9 gene in postnatal growth profoundly altered the cartilage homeostasis of the growth plate, the articular cartilage, and the IVD.

Subjects and Methods Mice All mice were maintained on a hybrid 129S7; C57 Bl6 genetic background. For most of the studies, male littermates of two different genotypes; namely, the control group Sox9flox/flox and the experimental group Agc1CreERT2; Sox9flox/flox were injected intraperitoneally with tamoxifen (T5648; Sigma, St. Louis, MO, USA) emulsified in sunflower seed oil (S5007; Sigma) at a concentration of 3 mg/10 g of mouse body weight. Mice were sacrificed according to institutional animal regulations.

Genotyping of mice See Supplemental Materials and Methods.

Cell death and proliferation Cell death was analyzed by the terminal deoxynucleotidyl transferase–mediated deoxyuridine triphosphate-biotin nick end labeling (TUNEL) assay method using the TdT-FragEL DNA Fragmentation Detection kit according to the manufacturer (QIA33; Calbiochem, San Diego, CA, USA). Cell proliferation was analyzed by bromodeoxyuridine staining of sections from mice that had been injected intraperitoneally with 5-bromo-20 deoxyuridine (BrdU; Invitrogen, Carlsbad, CA, USA) at a dose of 0.1 mL/10 g body weight. Mice were sacrificed 2 hours later post–BrdU injection and later serial paraffin sections of the tibial growth plate were stained for BrdU with the BrdU detection kit (Zymed/Invitrogen). Those cells that were BrdU-positive were counted in tibial growth plate sections in a region spanning 1000 nm across the width of the growth plate.

2512

HENRY ET AL.

RNA analysis and quantitative real-time PCR RNA was harvested from the lumbar IVDs or the articular cartilage of the tibial plateau with the TRIzol reagent (Invitrogen), treated with Turbo DNase (Ambion/Life Technologies, Carlsbad, CA, USA), and then reverse-transcribed to first-strand cDNA with the high-capacity cDNA reverse transcriptase kit (Applied Biosciences) according to the manufacturer’s instructions. The articular cartilage was dissected from the tibial plateau by a blunt cut with a scalpel blade from the tibia bone of several 8-week-old Sox9 depleted Sox9D/D mice and Sox9flox/flox littermate control mice that had been injected with tamoxifen at 7 weeks. The quantitative PCR experiments were conducted with fluorogenic 50 exonuclease assays with the TaqMan Gene Expression Master Mix with the following primer/probe sets shown in Supplemental Materials and Methods.

X-ray, micro-computed tomography, and histomorphometry X-rays of mouse skeletons were performed by Faxitron. Femurs were denuded of soft tissue and were fixed overnight in 10% formalin, and kept at 70% ethanol for micro–computed tomography (mCT) scanning using the explore GE Locus SP system (GE Healthcare). The mCT image was analyzed with the MicroView 2.2 Software package (GE Healthcare). Histomorphometry results were examined with the OsteoMeasure histomorphometry system (OsteoMetrics).

RNA in situ hybridization Five-micrometer (5-mm) paraffin sections were subjected to RNA in situ hybridization using the Digoxigenin RNA labeling kit (Roche-Applied-Biosciences, Indianapolis, IN, USA) according to manufacturer’s instructions. RNA-labeled deoxyuridine triphosphate probes for Aggrecan, Collagen 1a1, Collagen 2a1, Collagen 10a1, Indian Hedgehog, and Parathyroid hormone receptor protein were used as described.(4)

Immunohistochemistry and immunofluorescence The tissues were fixed with 10% formalin overnight, decalcified with either 0.5 M EDTA pH 8.0 for 1 week or Morse’s Solution for 2 days and then embedded in paraffin. Five-micrometer (5 mm) paraffin sections were subjected to epitope retrieval with either boiling citrate buffer pH 6.0 or enzymatic incubation with 0.1% trypsin, followed by a hyaluronidase treatment (2 mg/mL in phosphate buffer pH 5.5) (MP Biomedicals, Solon, OH, USA). Immunohistochemistry was performed with the SuperPicture Kit AEC Broad kit (Invitrogen) or the SuperPicture rabbit kit (Invitrogen) with the True Blue Peroxidase substrate (KPL, Gaithersburg, MD, USA). Primary antibodies used for immunohistochemistry and immunofluorescence are listed in Supplemental Materials and Methods.

RNA seq materials and methods One microgram (1 mg) of total RNA from the lumbar IVD was extracted using the TRIzol method from Sox9-depleted mice and control mice and then processed into double-stranded cDNA (ds-cDNA) using the NuGen Ovation RNA-Seq System (cat. #7100; Journal of Bone and Mineral Research

NuGEN Technologies, Inc., San Carlos, CA, USA). Sequencing libraries were prepared using 100 ng of picogreen-quantified ds-cDNA using the standard method for the Illumina Paired End DNA Sample Prep Kit (cat. #PE102-1002; Illumina, Inc., San Diego, CA, USA). The library size was 250 to 325 bp and was enriched with 12 cycles of amplification. Libraries were quantified with picogreen and checked for sizing and amount of adapter dimers using a Bioanalyzer High Sensitivity chip (Agilent Technologies, Santa Clara, CA, USA). The libraries were sequenced on an Illumina GAII as paired end 36 nucleotides. The sequences were aligned to the genome using Efficient Alignment of Nucleotide Databases (ELAND) pair.

Results To circumvent the perinatal lethality of heterozygous Sox9 mutant mice and to inactivate cartilage-specific genes in cartilaginous tissues postnatally, a genetic strategy in mice using an inducible, conditional approach was adopted. The injection of the inducer, tamoxifen, into newborn, 1-week-old, or 2-week-old Agc1CreERT2; Sox9flox/flox mice resulted in lethality presumably due to collapse of the trachea or failure of rib cage expansion. Consequently for these studies, the earliest time point for tamoxifen injection was 18 days postnatal. Phenotypic changes were observed in the postnatal Sox9-depleted or Sox9D/D mice in three different tissues: growth plate cartilage, articular cartilage, and the IVD of the Sox9D/D mice.

Growth plate phenotype The most noticeable gross morphological change to the skeleton upon the postnatal disruption of the Sox9 gene was the arrest of the longitudinal growth of the long bones and severe kyphosis of the spine. The faxitron of a 4-month-old Sox9-depleted Sox9D/D mouse injected with tamoxifen at 6 weeks revealed a smaller body size and cervical flexure of the spine compared to control Sox9flox/flox mice (Fig. 1A). Even though the Sox9 protein is not expressed in bone, mCT of femurs from these Sox9D/D mice revealed a severe reduction of trabecular bone in the epiphysis of the femur underneath the primary spongiosa (Fig. 1B, C). Furthermore, morphometric analysis (data not shown) of the primary spongiosa showed a reduction in trabecular thickness and in total number of trabecula for the Sox9-depleted mice compared to controls. In the Sox9D/D mouse, the number of osteoclasts along the bone surfaces was higher as well as the area of erosion surfaces along the bone surface for the Sox9D/D mouse. To investigate the stunted growth of Sox9-depleted mice, histological sections were performed on the knee joint that included the femoral and tibial growth plates. The growth plate of a 6-week-old Sox9D/D mouse that was injected previously with tamoxifen at 3 weeks was compressed compared to control Sox9flox/flox mouse (Fig. 2A, B). Upon depletion of Sox9, the number of cells in the growth plate was reduced, and the remaining cells were not arranged in a columnar orientation (Fig. 2D, F) that is characteristic of a control growth plate (Fig. 2C, E). Furthermore, many cells in the Sox9D/D growth plate had a flattened, spindle-shaped morphology that was different from Journal of Bone and Mineral Research

Fig. 1. Skeletal phenotype of mice in which the Sox9 gene is disrupted during postnatal growth. Faxitron X-ray and mCT of 4-month-old mice injected with tamoxifen at 6 weeks. (A) Left: X-ray of control Sox9flox/flox mouse; right: X-ray of Sox9D/D mouse. Arrow points to kyphosis of the cervical region of the spine. (B) mCT of Sox9flox/flox femur. (C) mCT of Sox9D/D femur. Arrow points to loss of trabecular bone in the epiphysis of femur underneath the primary spongiosa.

the morphology of the stacked proliferative chondrocytes in the Sox9flox/flox mice (Fig. 2D). Safranin-O staining, a vital stain for sulfated proteoglycans, was dramatically reduced in the Sox9depleted growth plate. The reduction of Safranin-O staining upon disruption of the Sox9 gene was observed initially in the proliferative zone, while Safranin-O was retained in the hypertrophic zone of the growth plate in Sox9D/D mice. The high THE POSTNATAL ROLE OF Sox9 IN CARTILAGE

2513

Fig. 2. Growth plate examined by histology, proliferation, and cell death assays in postnatal Sox9-depleted mice. Six-week-old tibial-femoral joint stained with Safranin-O/Fast Green from mice injected with tamoxifen at 3 weeks: (A) Sox9flox/flox, (B) Sox9D/D mouse, (C) Sox9flox/flox insert, (D) Sox9D/D mouse; some cells have flattened morphology, see insert. BrdU-stained section from 4-week-old mouse injected with tamoxifen at 3 weeks: (E) Sox9flox/flox insert, (F) Sox9D/D, arrow points to hypocellularity in prehypertrophic zone. Sections of growth plate stained with TUNEL: (G) Sox9flox/flox, (H) TUNEL-positive cells detected in Sox9D/D mouse. (I) Bar graph showing number of cells (n) on the y-axis in a 1000-mm width across the growth plate: BrdU, Sox9flox/flox n ¼ 114; Sox9D/D n ¼ 33; TUNEL, Sox9flox/flox n ¼ 0, Sox9D/D n ¼ 144 with standard deviation. (J,K) Indirect immunofluorescent staining of a 5-week-old growth plate section with antibody LC3B from mice injected with tamoxifen at 3 weeks. Arrow points to LC3B positive staining. Ectopic LC3B detected in growth plate (J). Inserts in panels C–H in lower left corner are magnifications of other box.

degree of disorganization and compression of the growth plate often precipitated its precocious closure. The proliferation rate of growth-plate chondrocytes was markedly reduced in the Sox9D/D mouse as shown by bromodeoxyuridine (BrdU; Fig. 2F, K). Also, the total number of cells occupying the proliferative zone was lower in the Sox9deficient mice, often with pockets of hypocellularity in the center of the growth plate above the hypertrophic zone of these mice (Fig. 2F). Furthermore, several cells in the Sox9-depleted growth plate were undergoing DNA fragmentation as detected by the terminal deoxynucleotidyl transferase–mediated deoxyuridine triphosphate-biotin nick end labeling (TUNEL; Fig. 2H, K). In addition, indirect immunofluorescence detected several cells in the center of the Sox9D/D growth plate that were positive for autophagy (Fig. 2J), whereas many fewer positive cells were seen in controls (Fig. 2I). Molecular analyses of the growth plate were performed by indirect immunofluorescence of sections from the growth plate of 25-day-old day animals that were injected previously with tamoxifen at 18 days. As expected, in control Sox9flox/flox mouse

2514

HENRY ET AL.

(Fig. 3A, C), Sox9 immunostaining was seen in the chondrocytes of the reserve, proliferative, and prehypertrophic zones, but no signal was detected in the hypertrophic zone. A dramatic reduction of Sox9 immunostaining was detected in the compressed Sox9D/D growth plate (Fig. 3B, D). The efficiency of the postnatal ablation of Sox9 in the growth plate of 25-day-old Sox9-depleted Sox9D/D mice injected with tamoxifen at 18 days was 85%. For the 7 days after the tamoxifen injection, the Collagen II (Fig. 3D) protein persisted in the extracellular matrix (ECM) of the growth plate in the absence of Sox9 nuclear staining (Fig. 3D). However, Collagen II (Fig. 3D) and Collagen IX (Fig. 3F) protein levels were reduced in the primary spongiosa of bone in the Sox9D/D growth plate. On the other hand, the Aggrecan (Fig. 3J) and Link (HAPLN1; Fig. 3L) proteins were largely absent from both the growth plate and primary spongiosa of Sox9D/D growth plate, demonstrating that the collagen proteins were more stable than aggrecan and link proteins upon the depletion of Sox9. The size of the hypertrophic zone was much smaller in the Sox9D/D growth plate, as shown by anti-Collagen Type X immunostaining (Fig. 3H). The ADAMTS5 Journal of Bone and Mineral Research

Fig. 3. Growth plate characterized with indirect immunofluorescence in Sox9-depleted mice. Growth plate of 25-day-old day animals that were injected previously with tamoxifen at 18 days. Sox9 immunostaining in Sox9flox/flox (A), and Sox9D/D (B). Merged dual immunofluorescence with nuclear staining Sox9 (red) and extracellular matrix (ECM) staining of Collagen II (green) in Sox9flox/flox (C) and Sox9D/D (D). Collagen IX immunostaining in Sox9flox/flox (E) and Sox9D/D (F). Merged dual immunofluorescence staining with a Collagen X antibody (red) and DAPI nuclear acid stain (blue) in Sox9flox/flox (G) and Sox9D/D (H). Aggrecan immunostaining in Sox9flox/flox (I) and Sox9D/D (J). Link protein immunostaining in Sox9flox/flox (K) and Sox9D/D (L). ADAMTS5 immunostaining in femoral growth plate with weak expression in reserve zone (arrow) in Sox9flox/flox (M) and stronger immunostaining in Sox9D/D (N). Histological sections from the growth plate of 21-day-old day animals that were injected previously with tamoxifen at 18 days. Dual indirect immunofluorescence immunostaining with a Collagen Type I antibody (red) and Collagen Type X antibody (green) in Sox9flox/flox (O) and Sox9D/D (P). Ectopic Collagen Type I protein is detected in ECM of prehypertrophic zone in Sox9D/D growth plate (arrow in P). po ¼ primary ossification; gp ¼ growth plate; so ¼ secondary ossification.

protein was expressed in the reserve zone chondrocytes of the control femoral growth plate with very weak expression in hypertrophic chondrocytes (Fig. 3M), but a much stronger immunostaining signal was detected in the reserve zone of the Sox9-deficient growth plate (Fig. 3N). Interestingly, Collagen I protein was detected, ectopically, in the ECM of the prehypertrophic zone just above the Collagen X–expressing hypertrophic zone in the Sox9D/D-deficient growth plate (Fig. 3P), and this expression pattern correlates with the Collagen 1a1 mRNA expression pattern (Supplemental Fig. 1D). It is plausible that the inactivation of Sox9 in the proliferative and prehypertrophic zones of the Sox9D/D growth plate leads to dedifferentiation of the chondrocytes. However, the bone-specific markers matrix metalloproteinase 13 (MMP13) and bone sialoprotein (BSP) were not detected in the prehypertrophic zone of the postnatal Sox9depleted growth plate at the time point tested (Supplemental Journal of Bone and Mineral Research

Fig. 2D, F). Nevertheless, the immunostaining intensity of MMP13 and BSP was stronger in the hypertrophic zone of the Sox9depleted growth plate compared to controls. Additional molecular analysis of several genes expressed during chondrocyte differentiation was performed by RNA in situ hybridization on histological sections. In sections from the tibialfemoral joint of 8-week-old Sox9flox/flox and Sox9D/D mice that were injected previously with tamoxifen at 7 weeks, the Collagen 2a1 (Col2a1) mRNA was dramatically reduced from growth plate and articular chondrocytes of the Sox9D/D mouse (Supplemental Fig. 1B). Aggrecan mRNA was also dramatically reduced in the proliferative zone of the growth plate in the Sox9D/D animals (Supplemental Fig. 1F). Two markers for the prehypertrophic zone, Indian Hedgehog (Ihh) (Supplemental Fig. 1J) and Parathyroid hormone related peptide receptor (Pth1R) (Supplemental Fig. 1L), as well as a marker for the hypertrophic zone, THE POSTNATAL ROLE OF Sox9 IN CARTILAGE

2515

Collagen 10a1 (Col10a1), were also reduced in the growth plates of Sox9D/D mice compared to Sox9flox/flox mice (Supplemental Fig. 1H).

Articular cartilage phenotype To examine the role of Sox9 in the articular cartilage, the Agc1CreERT2; Sox9flox/flox mice were injected with tamoxifen, starting at 6 weeks and sacrificed at later time points. The postnatal depletion of the Sox9 protein was very efficient, as determined by indirect immunofluorescence with a Sox9 antibody (92% ablation efficiency of Sox9 in 5-month-old Sox9-depleted Sox9D/D mice injected with tamoxifen at 2 months; Fig. 5B). Furthermore, the quantitative RT-PCR experiments revealed that the levels of Sox9 transcript were downregulated 8.2-fold in the articular cartilage of 8-week-old Sox9-depleted Sox9D/D mice compared to Sox9flox/flox control mice. One consequence of the deletion of Sox9 was the loss of

Safranin-O staining above the tidemark (Fig. 4B). In mice injected with tamoxifen at 6 weeks (Fig. 4B), but not in mice injected at 3 months or later (Fig. 4F, H), the articular cartilage of the Sox9D/D mice was thinner in these mice compared to control Sox9flox/flox mice. We speculate that this could be due to a slower development of the articular cartilage when Sox9 was deleted at an earlier stage. However, no histopathological signs of fibrillation or clefting at the articular surface of the knee joint were observed at the light microscopy level in 4-, 6-, or 12-month-old Sox9-depleted mice that were injected with tamoxifen starting at 6 weeks (n ¼ 40 mice). Furthermore, adult 18-month-old Sox9D/D mice injected with only one injection of tamoxifen at 6 months did not display hallmarks of premature osteoarthritis either despite the lack of Safranin-O staining (Fig. 4F). In these mice, Safranin-O staining was still maintained in the growth plate of the Sox9D/D mice (Fig. 4F). Aggrecan is an abundant sulfated proteoglycan in the articular cartilage, and its expression level correlates with the intensity of

Fig. 4. Articular cartilage stained with Safranin-O/Fast Green in postnatal Sox9-depleted mice. Sections from 4-month-old mice injected with tamoxifen at 6 weeks. (A) Sox9flox/flox; (B) Sox9D/D loss of Saf-O staining in uncalcified articular cartilage above the tidemark (arrow). Knee sections from 18-month-old mice injected with tamoxiphen at 6 months: (C) Sox9flox/flox; (D) magnified view of articular cartilage Sox9flox/flox; (E) Sox9D/D; (F) magnified view of articular cartilage Sox9D/D. Saf-O staining is reduced in the uncalcified articular cartilage above the tidemark (arrow).

2516

HENRY ET AL.

Journal of Bone and Mineral Research

Fig. 5. Articular cartilage characterized with immunohistochemistry and indirect immunofluorescence in postnatal Sox9-depleted mice. Sections of the articular cartilage from 5-month-old mice injected with tamoxifen at 2 months: (A) Sox9 immunostaining in uncalcified cartilage above the tidemark Sox9flox/flox; (B) loss of Sox9 immunostaining in articular cartilage Sox9D/D. Dual indirect immunofluorescence with a Sox9 antibody (red) and a Collagen Type II antibody (green) of Sox9flox/flox (C) and Sox9D/D mouse (D). Sections of the articular cartilage from 9-week-old mice injected with tamoxiphen at 8 weeks: (E) aggrecan immunostaining in the territorial ECM of Sox9flox/flox mouse; (F) aggrecan levels in the territorial matrix reduced, Sox9D/D. Sections of the knee joint from 5-month-old mice injected with tamoxiphen at 3 months: aggrecan immunostaining Sox9flox/flox (G) and Sox9D/D mouse (H), aggrecan is lost from the surface articular cartilage above the tidemark; ADAMTS5 immunostaining Sox9flox/flox (I) and stronger ADAMTS5 immunostaining Sox9D/D (J); and immunostaining with an anti-aggrecan neoepitope antibody, NITEGE, in Sox9flox/flox (K) and Sox9D/D (L). Sections of the knee joint from 5-month-old mice injected with tamoxifen at 3 months and subjected to indirect immunofluorescence with an anti-decorin antibody: (M) weak decorin immunostaining surface articular cartilage Sox9flox/flox mouse; (N) strong decorin immunostaining in Sox9D/D articular cartilage; lubricin immunostaining intensity is similar in Sox9flox/flox (O) and Sox9D/D (P).

Safranin-O vital staining.(14) In knee sections of 9-week-old Sox9D/D mice injected with tamoxifen at 8 weeks, the aggrecan levels in the territorial ECM of the articular cartilage (Fig. 5F) were reduced, but aggrecan immunostaining was still detected close to the surface of articular chondrocytes. A dramatic loss of aggrecan protein from the uncalcified articular cartilage was observed in knee sections from 5-month-old mice previously injected with tamoxifen at 3 months (Fig. 5H). Interestingly, the proteolytic enzyme, ADAMTS5,(15) which cleaves the aggrecan protein at multiple sites was detected at markedly higher levels in the Sox9D/D mouse compared to the Sox9flox/flox mouse (Fig. 5J). Furthermore, one of the cleavage product of aggrecan in the articular cartilage was detected with an aggrecan neoepitope antibody NITEGE(15–17) at higher levels in the surface articular cartilage of the Sox9D/D mouse (Fig. 5L). Journal of Bone and Mineral Research

The intensity of the immunofluorescent signal for the Collagen II protein in the surface articular cartilage of 5-month-old Sox9D/D mice previously injected with tamoxifen at 2 months was similar to the signal detected in control Sox9flox/flox mice, suggesting that inactivation of the Sox9 gene did not alter the Collagen II protein stability in the ECM of the articular cartilage (Fig. 5C, D). Nevertheless, the amount of Collagen II staining detected in the ossified secondary ossification center of the Sox9-deficient mice was reduced, implying that de novo synthesis of Collagen II protein ceased in the articular cartilage during the 3-month period or that Collagen II protein degradation was more rapid in the secondary ossification center of Sox9D/D mouse compared to the Sox9flox/flox mouse. Likewise, the levels of Lubricin protein (PRG4), a secreted protein whose expression is restricted to the surface articular cartilage and is not present in growth plate, did THE POSTNATAL ROLE OF Sox9 IN CARTILAGE

2517

not change upon depletion of Sox9 in 5-month-old mice injected with tamoxifen at 3 months (Fig. 5O, P). However, the intensity of immunostaining for Decorin was higher in the articular cartilage of Sox9-depleted mice (Fig. 5N) compared to control Sox9flox/flox mice.

IVD phenotype Although the depletion of Sox9 leads to severe kyphosis in the growing postnatal skeleton (Fig. 1A), the Sox9 gene continues to play a role in adult mouse spine homeostasis. Sox9 is expressed in the notochord-derived cells of the nucleus pulposus and in the sclerotome-derived fibrocartilage cells of the annulus fibrosis of the IVD (Fig. 7A). Sox9 is also expressed in the growth plate chondrocytes of the vertebral body. The most apparent phenotype upon X-ray of 6-month-old spines from Sox9D/D mice injected with tamoxifen 2 months earlier was the compression of the IVDs (Fig. 6A). This compression is more clearly seen by mCT as shown in the IVD between lumbar vertebrae 3 and 4 (L3 and L4; Fig. 6C) and was also observed in the other lumbar disks and the thoracic and cervical IVDs (data not shown). The loss of Sox9 in the growth plate of the vertebrae leads to precocious growth plate closure (Fig. 6C, E), although this does not affect the size of individual vertebrae since adult

4-month-old animals exhibit very little additional growth in the axial skeleton or the limbs. The growth plate closure phenotype is also shown with histological Von Kossa staining of the lumbar 4/5 vertebrae (Fig. 6E). Collagen II immunostaining demonstrated that ECM Collagen II protein was detected in the annulus fibrosis, but not the nucleus pulposus of the IVD (Fig. 7B), and the efficiency of the postnatal ablation of Sox9 in the IVD of 3-month-old Sox9depleted Sox9D/D mice injected with tamoxifen at 2 months old was 94%. The depletion of Sox9 from the IVD in 10-week-old mice that were injected with tamoxifen at 2 weeks old resulted in a dramatic loss of Safranin-O staining from the nucleus pulposus (Fig. 7E, F). In other Sox9D/D mice, Safranin-O staining was very low in the nucleus pulposus and very sparse in the annulus fibrosis, but was still readily detectable in the growth plate, suggesting that aggrecan loss occurs more rapidly in the IVD compared to the vertebral growth plate (Fig. 7G, H). The diminution of aggrecan immunohistochemical staining in the IVD (Fig. 7I, J) correlates with the reduction of Safranin O staining in the Sox9D/D mice. The depletion of Sox9 in 8-month-old Sox9D/D mice injected with tamoxifen at 6 months old resulted in severe disk compression (Fig. 7K, L). We then performed quantitative RT-PCR experiments (qPCR) of several genes expressed in the IVD (Supplemental

Fig. 6. Vertebral column of adult Sox9-depleted mice. Spines from 6-month-old animals injected with tamoxifen at 4 months. (A) Faxitron of Sox9flox/flox mouse (left) and Sox9D/D mouse (right). Arrow points to an empty space visualized in the X-ray, and presumably the intervertebral disc (IVD). mCT of the lumbar vertebrae 3 and 4 (L3/L4) in the Sox9flox/flox mouse (B) and Sox9D/D mouse (C). The IVD (L3/L4) is depicted as the empty space shown with a red arrow (top) and white arrow (bottom). Compression of IVD Sox9D/D: (D) Von Kossa staining of L4/L5 vertebra in Sox9flox/flox mouse, black arrows point to vertebrae growth plate; (E) Von Kossa staining of lumbar 4/5 vertebra in Sox9D/D mouse.

2518

HENRY ET AL.

Journal of Bone and Mineral Research

Fig. 7. Histological sections of lumbar intervertebral disc (IVD). IVD from 3-month-old injected with tamoxiphen at 2 months old: (A) Indirect immunofluorescence with anti-Sox9 antibody Sox9flox/flox; (B) Sox9 antibody (red nuclear) and Collagen Type II antibody (green ECM) of Sox9flox/flox; (C) Sox9 immunostaining, Sox9D/D; (D) Sox9 antibody (red) and Collagen Type II antibody (green), Sox9D/D mouse. Saf-O/Fast Green sections of 10-weekold injected with tamoxiphen at 8 weeks old: (E) Sox9flox/flox; (F) Sox9D/D. Saf-O/Fast Green sections of 12-week-old injected with tamoxiphen at 8 weeks old: (G) Sox9flox/flox; (H) Sox9D/D. Immunohistochemistry with anti-aggrecan antibody of 10-week-old mice injected with tamoxiphen at 8 weeks old: (I) Sox9flox/flox; (J) Sox9D/D Aggrecan depletion in IVD, but aggrecan detected in growth plate, correlating with the Saf-O levels observed in G and H. Saf-O/Fast Green sections of 8-month-old mice injected with tamoxiphen at 6 months old: (K) Sox9flox/flox; (L) Sox9D/D.

Fig. 3) of 8-week-old Sox9D/D mice and control mice that had been injected with tamoxifen at 7 weeks old. The Sox9 mRNA was downregulated 9.6-fold in the Sox9D/D mice. The ECM genes aggrecan (19.8-fold reduction) and HAPLN1 (14.6-fold reduction) were the most dramatically downregulated genes of those tested. The mRNA for the MMP genes, MMP3 and MMP9, were upregulated (þ2.5-fold and þ1.9-fold, respectively), as well as the ADAMTS5 gene (þ1.3-fold) in the Sox9D/D mice, suggesting that several ECM components might be degraded. To further characterize the molecular changes resulting from Sox9 depletion in the IVD, we analyzed the global RNA profile from the IVD of 6-month-old Sox9D/D and control animals that had been injected with tamoxifen at 5 months old with wholetranscriptome shotgun sequencing technology or RNA-Seq. A list of differentially expressed genes upon the depletion of Sox9 is presented in Table 1. The most downregulated gene was Collagen 10a1, presumably an indirect effect because Sox9 is not expressed in hypertrophic chondrocytes. Some of the downregulated genes have been previously described as direct transcriptional targets of Sox9.(18) The mRNAs for several extracellular proteins, including those of the three polypeptide chains encoding Collagen IX, as well as Collagen 2a1, Matrillin3, and upper zone of growth plate and cartilage matrix–associated (UCMA)(19) were among the 14 most downregulated. Transcripts Journal of Bone and Mineral Research

of several secreted molecules were downregulated such as secreted frizzled-related protein 5 (Sfrp5) and gremlin (Grem1), which are antagonists of the Wnt/b-catenin and bone morphogenic protein (BMP) signaling pathway, respectively. Furthermore, among the 14 most downregulated genes upon the depletion of Sox9, several are secreted molecules, including, Rspo4, Sfrp5, cytl1, and Gremlin. Sox9 has been reported to increase the BMP-antagonist gremlin expression in the pyloric epithelium(20) and the related BMP-antagonist Noggin has been reported to be a transcriptional target of Sox9.(4,18,21) Other downregulated cytokines include the molecule Rspo4, a member of the R-spondin family implicated in the Wnt/b-catenin signaling cascade,(22,23) and cytokine like molecule 1 (Cytl1), which is highly expressed in the prehypertrophic zone.(24) In addition, mRNAs encoding for cell-surface molecules were also among the 14 most downregulated such as Clec3a, a C-type lectin cell surface receptor protein with a strong expression in cartilage,(25) and Fxyd2, an enzyme subunit of the Na, K-ATPase ion channel reported to be an osmotic pump regulator in the kidney tubules.(26) Other genes encoding proteins localized to the ECM that were not represented in the list of the 14 most downregulated genes, include Cilp2 (log2 ¼ 2.8), Col27a1 (2.5), Thbs1 (2.3), Chad (2.2), Hapln1 (2.1), Col11a2 (2.1), Comp (1.9), Vcan (1.8), THE POSTNATAL ROLE OF Sox9 IN CARTILAGE

2519

Table 1. Fourteen of the Most Downregulated and Upregulated Genes Upon the Postnatal Depletion of Sox9 in the Adult Mouse Intervertebral Disk Gene symbol

log2 (fold change)

Fold change

4.83 4.53 4.20 3.91 3.74 3.46 3.19 2.98 2.97 2.83 2.79 2.77 2.71 2.70 2.59

23.31 20.48 17.62 15.27 13.95 11.97 10.16 8.89 8.82 7.99 7.81 7.67 7.36 7.31 6.70

6.49 5.26 5.13 5.07 4.79 4.52 4.47 4.37 4.32 4.30 3.99 3.99 3.79 3.67

42.12 27.66’ 26.35 25.69 22.95 20.45 19.94 19.06 18.63 18.50 15.94 15.91 14.34 13.47

Downregulated Col10a1 Col9al Matn3 Ucma Rspo4 Col9a3 AA465934 Col2a1 Col9a2 Clec3a Sfrp5 Fxyd2 Cytl1 Grem1 Ppp1r1b Upregulated Dbh Cadps Syt1 Ucp1 Syt4 Scg2 Rab3c Ddc Mpo Retnig Elane Cxcr2 Slc27a2 Ppbp

Cilp (1.6), Acan (1.4), and Bgn (0.80). A few cytokines were also downregulated, such as chondromodulin (Lect1, 2.1), and FrzB (encoding the secreted frizzled related receptor 3; 2.7), and insulin-like growth factor binding protein (Igfbp5) (1.5). Several mRNAs encoding enzymes were downregulated, including lysyl-like oxidase 2 (loxl2; 1.3), lysyl-like oxidase 3 (loxl3; 1.1), and lysyl-like oxidase 4 (loxl4; 1.6), responsible for collagen cross-linking,(27) and 30 phosphoadenosine 50 phosphosulfate transferase (papss2; 1.9), a key enzyme for sulfation of proteoglycans.(28) Several cell-surface receptors were downregulated, including the mechanoreceptor (Fam38b or Piezo2; 1.6),(29) and cation ion channel/receptor polycystin-1 (Pkd1; 1.2),(30) fibroblast growth factor receptor 3 (FGRR3; 1.3), the osmotic pressure transient potential cation channel subfamily V member 4 (TRPV4; 1.5),(31,32) and G-coupled receptor 126 (GPR126; 1.4).(33) A pie chart depicting the subcellular localization of 259 downregulated genes in the Sox9-depleted IVD illustrates that genes encoding for proteins localized to the extracellular space

2520

HENRY ET AL.

are differentially expressed at the highest frequency (Supplemental Fig. 4). Nevertheless, genes encoding for proteins localized to the plasma membrane were also differentially expressed at a high frequency, including nine different G-coupled receptors, 13 transmembrane receptors, and nine ion channels. Moreover, genes encoding for 37 different enzymes were also downregulated. The table of the 14 most upregulated genes upon the postnatal depletion of Sox9 in the adult IVD revealed two trends (Table 1). First, several genes involved in sympathetic neural transmission, such as dopamine beta-hydroxylase (Dbh) and dopamine decarboxylase (Ddc), were upregulated. Also, other genes encoding membrane-trafficking proteins committed to vesicular transport in neurons such the synaptotagmins, Syt1 and Syt4, were also upregulated. Second, several genes implicated in the inflammatory response, particular to reactive oxygen species, were also upregulated, such as thermogenin (Ucp1), myeloperoxidase (Mpo), and the elastase neutrophil-expressed molecules (Elane). Interestingly, the chemokine CXCR2, a chemokine implicated in the activation of the apoptotic pathway in chondrocytes, was also upregulated.(34) The list of downregulated genes from the RNA-Seq with a Z-score of at least 10 was submitted to the Database for Annotation, Visualization, and Integrated Discovery (DAVID) Bioinformatics server at the National Institute of Allergy and Infectious Diseases (NIAID) division of the NIH.(35) The Gene Ontology classification categories (http://www.geneontology.org) for the downregulated genes with the lowest log(p) were cell adhesion and skeletal system development (Supplemental Table 1, top); whereas the Gene Ontology classification for the upregulated genes with the lowest log(p) were the generation of precursor metabolites and energy as well aerobic respiration and oxidative reduction (Supplemental Table 1, bottom). Likewise, the same list of downregulated genes was submitted to the Ingenuity Systems Software Package (http://www.ingenuity.com), and the disease-related genes with the lowest log(p) were connective tissue disorders and skeletal/muscular disorders (Supplemental Fig. 5). Additional disease gene analysis was performed using the DAVID program, and the diseases associated with the downregulated genes were annotated according to categories with the lowest log(p). These diseases included skeletal dysplasias, connective tissue disorders, IVD disorders, and arthritis (Supplemental Table 2). Because we did not enrich for polyadenylated genes in our RNA purification scheme and used total RNA, we were able to obtain a list of differentially expressed long noncoding RNA (ncRNA) upon depletion of Sox9 in the IVD from RNA-Seq experiments (Supplemental Tables 3 and 4). Interestingly, many of the downregulated long ncRNA molecules were embedded in the intronic regions of genes whose mRNA was also downregulated. Long ncRNAs were located in the introns of the enzymes Loxl2, a lysyl oxidase, and Papss2, a phosphosulfate synthetase. Another ncRNA was embedded in the intron of Hapln1 (encoding link protein) and the transcription factor Sox5. Furthermore, two other ncRNA downregulated upon the depletion of Sox9 were located near Tcf7l2 and Pik3r1, molecules involved in glucose metabolism. Two independent ncRNAs were located in the first intron of the signaling molecule Smad3. Journal of Bone and Mineral Research

Additionally, several ncRNA were also upregulated upon the depletion of Sox9 in the IVD. Again, some of these ncRNAs were embedded in the intronic region of genes implicated in gluconeogenesis and carbohydrate metabolism.

Discussion The transcription factor Sox9 continues to play a major role in the homeostasis of cartilage in postnatal and adult mice. The consequences of the inducible, conditional disruption of the Sox9 gene during the period of postnatal growth or in adult mice are different from those of a germline mutation. Nevertheless, the deletion of the Sox9 gene starting at 6 weeks old led to stunted growth and kyphosis of the spinal column, thereby evoking some of the skeletal phenotypes manifested in the germline disease campomelic dysplasia.(1,2) The growth arrest in the postnatal Sox9-deficient mice is due to a malfunctioning growth plate characterized by lower proliferation and increased cell death of growth plate chondrocytes. The detection of Col1a1 mRNA-positive cells and extracellular localized Collagen Type I protein in the prehypertrophic zone of the Sox9-deficient growth plate suggests that some Sox9deficient chondrocytes adopt a different cell fate or become dedifferentiated. However, neither BSP or MMP13 protein was ectopically expressed in the prehypertrophic zones of the Sox9depleted mice, although the signal of immunofluorescence was stronger in the hypertrophic zone of the Sox9-depleted mice. The absence of BSP and MMP13 in cells of the prehypertrophic zone, where Collagen I is ectopically expressed, would suggest that at the postnatal time point tested, these cells have not yet adopted an osteoblastic lineage fate. The cell death observed 7 days after tamoxifen injection maybe a direct consequence of the loss of Sox9 transcriptional activity if certain transcriptional targets were critical for chondrocyte survival in the growth plate. Additionally, the loss of expression of the transcriptional targets of Sox9 (ie, secreted signaling molecules, ECM proteins, and others) may create an environment that is not permissive for cell survival. The tidemark of adult articular cartilage is a basophilic line that delineates the boundary between the surface uncalcified cartilage and deeper calcified cartilage. The inducible cre recombinase activity of the Agc1CreERT2 mouse line is restricted to the articular chondrocytes of the upper uncalcified cartilage above the tidemark.(8) Furthermore, Sox9 is almost exclusively localized to the articular chondrocytes of the uncalcified cartilage, with very few Sox9-positive cells in the region of calcified cartilage.(36) The depletion of Sox9 in the articular chondrocytes of the uncalcified zone led to the loss of Safranin-O staining and aggrecan immunostaining in this zone. The absence of aggrecan mRNA in the articular cartilage (data not shown), similar to the growth plate, suggests that the disruption of the Sox9 gene postnatally halts the synthesis of de novo aggrecan protein. On the other hand, we also hypothesize that the degradation of the aggrecan protein is also accelerated in the uncalcified articular cartilage upon the depletion of Sox9. The reduction of aggrecan in the territorial matrix of the articular cartilage in Sox9D/D mice was also accompanied by higher Journal of Bone and Mineral Research

levels of ADAMTS5 enzyme and increased NITEGE aggrecan neoepitope. Moreover, increased expression of ADAMTS5 in the Sox9D/D mice was also observed in two other Sox9-depleted tissues—the reserve zone of the growth plate and the IVD. Chromatin immunoprecipitation experiments revealed that Sox9 interaction sites were identified in both the Aggrecan(37) and ADAMTS5 genes of rat chondrosarcoma cells,(18) suggesting that Sox9 transcriptional activity may orchestrate the synthesis and degradation of the aggrecan protein in a coordinated fashion. Histopathological signs of premature osteoarthritis, such as fibrillation and clefting of the articular cartilage, were not observed in Sox9D/D mice even though sulfated proteoglycans in the uncalcified cartilage were absent for a period up to 1 year. In contrast, the Collagen II protein was detected by immunostaining in the uncalcified articular cartilage of the Sox9-depleted mice in spite of the severe loss of Collagen 2a1 mRNA, suggesting that the Collagen II fibril network was maintained in the articular cartilage of the Sox9D/D mice. One explanation for this phenomenon could be that the cross-linking of the Collagen II fibrils may have occurred prior to tamoxifen injection, thereby stabilizing the collagen fibril network. The reason for the absence of osteoarthritis might be at least in part accounted for by the quadripedic anatomy of mice, in which weight bearing is distributed over four limbs, and by the relative sedentary nature of mice in laboratory cages. A surgical intervention in the joint of Sox9D/D mice may lead to more obvious signs of osteoarthritis. Most mouse models for OA mutations are germline mutations, whereas our gene disruption is induced postnatally. Although we have shown that the postnatal loss of Sox9 in adult mouse articular cartilage does not result in overt, clinical osteoarthritis, others have shown that the conditional activation (gain-of-function) of a constitutively active b-Catenin in adult mice by tamoxifen injection of adult Col2a1-CreERT2; b-Cateninfx(Ex3)/wt mice leads to an osteoarthritic phenotype with severe erosion of the articular cartilage accompanied by accelerated chondrocyte differentiation with the upregulation of Collagen X mRNA and increased MMP13 immunostaining in the articular cartilage.(38) In another study, b-Catenin was only transiently (not permanently) activated in cartilage starting at two weeks postnatal, and the transgenic mice were later euthanized at 3 weeks of age.(39) The articular cartilage of these transgenic mice had an acute loss of proteoglycan and accelerated aggrecan degradation closely resembling those observed in our Sox9D/D mice. In reciprocal experiments, b-Catenin was permanently deleted in 5-day-old Col2a1-CreERT;b-Cateninfx/fx mice before the development of the articular cartilage, and these mice were euthanized later at 7 weeks of age. These mice exhibited clefting and deterioration of the articular cartilage; however, it is difficult to compare these studies to our postnatal Sox9 depletion studies, since our experiments were performed starting at 3 weeks postnatally or later in adult mice after the establishment of the articular cartilage. Furthermore, the IVD was also studied in these transgenic mice in which b-Catenin was transiently activated (gain-of-function) or permanently disrupted (loss-of–function).(40) Although the Sox9-depleted mice do not display overt osteoarthritis, the drastic proteoglycan reduction in the uncalcified zone of the articular cartilage was not replenished in adult THE POSTNATAL ROLE OF Sox9 IN CARTILAGE

2521

mice after the Sox9 gene was disrupted in the articular chondrocytes of the uncalcified cartilage. In adult 18-monthold Sox9D/D mice that had been injected with one injection of tamoxifen at 6 months, the upper uncalcified articular cartilage was still devoid of Safranin-O staining in the uncalcified articular cartilage. The complete lack of de novo proteoglycan synthesis in the 12-month period points to either an inability of the highly quiescent articular chondrocytes to mount a repair process in the uncalcified articular cartilage in the absence of Sox9 or that the trigger for repair was insufficient. The presence of Safranin-O staining in the deeper calcified cartilage of the Sox9-depleted mice may be due to preexisting aggrecan protein in the ECM of the calcified zone prior to tamoxifen injection that remained stable in the calcified cartilage. Nevertheless, chromatin immunoprecipitation experiments revealed that several ECM genes, such as Prg4 (encoding the lubricin protein) and Decorin, did not harbor Sox9 interaction sites in rat chondrosarcoma cells between 15 kb 50 to the start of transcription and þ10 kb 30 to the polyA site in these genes.(18) This would suggest that certain cartilage ECM genes may not be direct transcriptional targets of Sox9. In the Sox9D/D mice, lubricin protein was detected and decorin immunostaining was more robust in the articular cartilage, suggesting that some genetic compensatory mechanisms might be operating in the Sox9-deficient mice. The homeostasis of the IVD was severely disrupted upon the depletion of Sox9. A dramatic loss of Safranin-O staining was observed in histology of the IVD 2 weeks after tamoxifen injection. Quantitative RT-PCR experiments with preselected genes revealed that several genes encoding for ECM components were downregulated upon the depletion of Sox9. Furthermore, enzymes responsible for the cleavage of ECM components, such as MMP3 and ADAMTS5, were also modestly upregulated in the Sox9-depleted mouse. For this qRT-PCR experiment, tamoxifen was injected postnatally in growing 7week-old mice. To better understand the postnatal role of Sox9 in the IVD of adult mice, we injected our mice at 5 months of age and waited 1 month postinjection to euthanize the mice and harvested total RNA from the lumbar IVDs. We obtained a global representation of genes that were differentially expressed upon the depletion of Sox9 in the IVD by RNA sequence technology. Several ECM genes were downregulated in the Sox9D/D IVD and, not surprisingly, many of these ECM genes have also been determined to be direct transcriptional targets of Sox9 in chondrocytic cells.(18) Interestingly, several of the most downregulated genes have been implicated in degenerative disk disease by genomewide association studies (GWAS) such as Collagen 9a1, Collagen 9a2,(41–44) aggrecan,(45–48) and CILP.(49) Additionally, several enzymes were also downregulated upon the depletion of Sox9 in the IVD. For example, three genes encoding for three different lysyl oxidase enzymes implicated in the biosynthetic cross-linking of collagen polypeptide chains at lysine residues (lysyl oxidases like 2, 3, and 4)(50) were all downregulated. Furthermore, another gene encoding for the obligate sulfotransferase enzyme, papps2 encoding for 30 phosphoadenosine 50 phosphosulfate transferase responsible for the posttranslational sulfation of proteins such as proteoglycans, as well as carbohydrates and lipids, was downregulated in

2522

HENRY ET AL.

the IVD. Two lines of evidence support that the papss2 gene plays an important role in skeletogenesis. The autosomal recessive mutation of the mouse papps2 gene results in the brachymorph mouse (Bm)(51) that exhibits shorter limbs, and a mutation in the orthologous human gene results in Pakistani type of human spondyloepimetaphyseal dysplasia.(52) Moreover, the papss2 gene was identified as the most differentially expressed enzyme gene in a microarray experiment using limb buds from embryonic day 12.5 (E12.5) Sox9-30 enhanced green fluorescent protein (EGFP) knock-in embryos compared to limb buds from E12.5 Sox9-EGFP/EGFP null chimeric embryos,(53) suggesting that papps2 may be a transcriptional target of Sox9. We speculate that a Sox9 transcription factor may have a bifunctional role to coordinate the synthesis, as well as posttranslational modification of several ECM proteins. The loss of proteoglycan, in particular aggrecan, was observed histologically in cartilaginous tissues of the Sox9-depleted mouse, including the growth plate, articular cartilage, and IVD. Aggrecan is the most abundant proteoglycan in these tissues, and this highly negatively-charged proteoglycan attracts water, and this osmotic property plays a major role in the compressibility and mechanical loading of the IVD(54) and the articular cartilage. Interestingly, the normal homeostasis of the IVD in the mouse is dependent on the levels of the aggrecan proteoglycan contained in the disk, because the heterozygote cartilage matrix deficient mouse (cmd/þ) containing only one functional copy of the mouse aggrecan gene exhibits IVD compression and degeneration at 1 year of age, but does not display abnormalities in the joints.(55) Consequently, in the mouse, the haploinsufficiency of the aggrecan gene has a direct impact upon IVD homeostasis. Therefore, we postulated that the loss of proteoglycan in the Sox9-depleted mice would diminish the hydration of the IVD, thus perturbing the mechanical loading of the disk, resulting in disk compression and subsequent degeneration. Upon examination of the list of 259 genes downregulated in the IVD of the Sox9-depleted mouse, several genes implicated in mechanotransduction were identified. For example, the polycystin gene (pkd1) that encodes for an integral membrane protein that has been shown to be a mechanical flow ‘‘sensor’’ in the renal epithelium was downregulated. The homozygous pkd1-null mice display skeletal abnormalities, including stunted growth and kyphosis of the spine.(30) Furthermore, the conditional inactivation of the pkd1 gene with a wnt1-cre driver results in abnormal mid-palatal suture expansion, suggesting that the pkd1 gene is important for mechanotransduction.(56) Another strong candidate was the gene TRPV4, encoding for the transient receptor potential villanoid calcium ion channel that is believed to be an ‘‘osmolarity sensor.’’(57,58) Human skeletal genetic disease associated with TRPV4 have been identified—the gain of function of TRPV4 leads to autosomal dominant brachyolmia(59) —whereas the loss of function of TRPV4 leads to inherited arthropathy.(60) The recently identified mechanoactivated ion channel encoded by the gene Piezo2 or Fam38b(29) was also downregulated in the IVD of the Sox9-depleted mouse. Hence, three genes implicated with mechanotransduction are downregulated upon depletion of Sox9 in the IVD. The downregulation of the G-coupled receptor 126 (GPR126) gene is Journal of Bone and Mineral Research

particularly interesting because genetic polymorphisms at the GPR126 locus may be associated with heritability of human height, as shown by three independent GWAS.(61–63) Furthermore, in one of the GWAS, the GPR126 association to overall human height was more specific to spine length (trunk) than to femur length (limb) or hip length. It would be very informative to identify Sox9 interaction sites in the chromatin environment of the IVD genome by chromatin immunoprecipitation sequence technologies (ChIP Seq) and compare these results with differential RNA sequence technology. The upregulation of many genes with high expression in the central nervous system upon depletion of Sox9 may be an indirect effect. It has been shown that the abnormal nerve ingrowth to the inner portion of the annulus fibrosis and the nucleus pulposus occurs in the diseased IVD.(64) Interestingly, sometimes this pathological innervation to the center of the diseased disk was sometimes accompanied by new blood vessel formation.(65) Another unexpected finding was the upregulation of many enzymes involved in aerobic respiration. We speculate that the hypoxic environment of the IVD is altered during the 1-month period of Sox9 depletion and that perhaps oxidative phosphorylation becomes more prominent. However, the compression of the IVD may lead to increased apoptosis that may be accompanied by an increase of reactive oxidative species rather than neovascularization of the inner IVD. Another advantage of using RNA sequence technology was the identification of many long ncRNAs that were either downregulated or upregulated in the IVD of the Sox9-depleted mouse. However, it is not known if these ncRNAs are direct transcriptional targets of Sox9. In conclusion, our results indicate that the role of Sox9 as a master differentiation factor, which has been previously documented during embryonic development of the skeleton, persists postnatally and in adult mice to control the homeostasis of connective tissues such as the growth plate, articular cartilage, and the IVD. In the IVDs, Sox9 coordinates, not only the expression of ECM genes, but also the expression of genes that have a role in the posttranslational modification of these ECM proteins, the genes for cytokines, cell-surface receptors, and ion channels that control the physiology of the cells in this tissue. We hypothesis that Sox9 may have a similar function of coordination in other cartilaginous tissues. We speculate that Sox9 could play a role in the pathology of degenerative diseases affecting cartilage, and that the molecules that control the expression or activity of Sox9 should be considered in therapeutic approaches to these diseases.

grants from the Arthritis Foundation (to SPH), the National Institutes of Health (NIH) T32 Training Grant (to SPH); and the NIH (AR053568 to BdeC). Authors’ roles: Study design: SPH and BdeC. Study conduct: SPH. Data collection: SPH and SL. Data analysis: SPH, SL, and KCA. Data interpretation: SPH and BdeC. Drafting manuscript: SPH and BdeC. Revising manuscript: SPH and BdeC. Approving final version of manuscript: SPH and BdeC. SPH takes responsibility for the integrity of the data analysis.

References 1. Foster JW, Dominguez-Steglich MA, Guioli S, Kwok C, Weller PA, Stevanovic´ M, Weissenbach J, Mansour S, Young ID, Goodfellow PN, Brook JD, Schafer AJ. Campomelic dysplasia and autosomal sex reversal caused by mutations in an SRY-related gene. Nature. 1994;372(6506):525–30. 2. Wagner T, Wirth J, Meyer J, Zabel B, Held M, Zimmer J, Pasantes J, Bricarelli FD, Keutel J, Hustert E, Wolf U, Tommerup N, Schempp W, Scherer G. Autosomal sex reversal and campomelic dysplasia are caused by mutations in and around the SRY-related gene SOX9. Cell. 1994;79(6):1111–20. 3. Bi W, Deng JM, Zhang Z, Behringer RR, de Crombrugghe B. Sox9 is required for cartilage formation. Nat Genet. 1999;22(1):85–9. 4. Akiyama H, Chaboissier MC, Martin JF, Schedl A, de Crombrugghe B. The transcription factor Sox9 has essential roles in successive steps of the chondrocyte differentiation pathway and is required for expression of Sox5 and Sox6. Genes Dev. 2002;16(21):2813–28. 5. Nakamura E, Nguyen MT, Mackem S. Kinetics of tamoxifen-regulated Cre activity in mice using a cartilage-specific CreER(T) to assay temporal activity windows along the proximodistal limb skeleton. Dev Dyn. 2006;235(9):2603–12. 6. Chen M, Lichtler AC, Sheu TJ, Xie C, Zhang X, O’Keefe RJ, Chen D. Generation of a transgenic mouse model with chondrocyte-specific and tamoxifen-inducible expression of Cre recombinase. Genesis. 2007;45(1):44–50. 7. Grover J, Roughley PJ. Generation of a transgenic mouse in which Cre recombinase is expressed under control of the type II collagen promoter and doxycycline administration. Matrix Biol. 2006;25(3): 158–65. 8. Henry SP, Jang CW, Deng JM, Zhang Z, Behringer RR, de Crombrugghe B. Generation of aggrecan-CreERT2 knockin mice for inducible Cre activity in adult cartilage. Genesis. 2009;47(12): 805–14. 9. Salminen H, Vuorio E, Sa¨a¨ma¨nen AM. Expression of Sox9 and type IIA procollagen during attempted repair of articular cartilage damage in a transgenic mouse model of osteoarthritis. Arthritis Rheum. 2001;44(4): 947–55. 10. Haag J, Gebhard PM, Aigner T. SOX gene expression in human osteoarthritic cartilage. Pathobiology. 2008;75(3):195–9.

Disclosures

11. Sive JI, Baird P, Jeziorsk M, Watkins A, Hoyland JA, Freemont AJ. Expression of chondrocyte markers by cells of normal and degenerate intervertebral discs. Mol Pathol. 2002;55(2):91–7.

All authors state that they have no conflicts of interest.

12. Barrionuevo F, Taketo MM, Scherer G, Kispert A. Sox9 is required for notochord maintenance in mice. Dev Biol. 2006;295(1):128–40.

Acknowledgments

13. Gruber HE, Norton HJ, Ingram JA, Hanley EN Jr. The SOX9 transcription factor in the human disc: decreased immunolocalization with age and disc degeneration. Spine (Phila Pa 1976). 2005;30(6): 625–30.

We thank Hongli Tang, MD, and Louis S Ramagli, PhD, for technical support from the DNA Analysis Core Facility at University of Texas MD Anderson Cancer Center for gene expression and RNA sequencing experiments. This work was supported by Journal of Bone and Mineral Research

14. Thalmann I, Machiki K, Calabro A, Hascall VC, Thalmann R. Uronic acid-containing glycosaminoglycans and keratan sulfate are present in the tectorial membrane of the inner ear: functional implications. Arch Biochem Biophys. 1993;307(2):391–6.

THE POSTNATAL ROLE OF Sox9 IN CARTILAGE

2523

15. Abbaszade I, Liu RQ, Yang F, Rosenfeld SA, Ross OH, Link JR, Ellis DM, Tortorella MD, Pratta MA, Hollis JM, Wynn R, Duke JL, George HJ, Hillman MC Jr, Murphy K, Wiswall BH, Copeland RA, Decicco CP, Bruckner R, Nagase H, Itoh Y, Newton RC, Magolda RL, Trzaskos JM, Hollis GF, Arner EC, Burn TC. Cloning and characterization of ADAMTS11, an aggrecanase from the ADAMTS family. J Biol Chem. 1999;274(33):23443–50. 16. Singer II, Scott S, Kawka DW, Bayne EK, Weidner JR, Williams HR, Mumford RA, Lark MW, McDonnell J, Christen AJ, Moore VL, Mudgett JS, Visco DM. Aggrecanase and metalloproteinase-specific aggrecan neo-epitopes are induced in the articular cartilage of mice with collagen II-induced arthritis. Osteoarthritis Cartilage. 1997;5(6): 407–18. 17. Gendron C, Kashiwagi M, Lim NH, Enghild JJ, Thøgersen IB, Hughes C, Caterson B, Nagase H. Proteolytic activities of human ADAMTS-5: comparative studies with ADAMTS-4. J Biol Chem. 2007;282(25): 18294–306. 18. Oh CD, Maity SN, Lu JF, Zhang J, Liang S, Coustry F, de Crombrugghe B, Yasuda H. Identification of SOX9 interaction sites in the genome of chondrocytes. PLoS One. 2010;5(4):e10113. 19. Surmann-Schmitt C, Dietz U, Kireva T, Adam N, Park J, Tagariello A, Onnerfjord P, Heinega˚rd D, Schlo¨tzer-Schrehardt U, Deutzmann R, von der Mark K, Stock M. Ucma, a novel secreted cartilage-specific protein with implications in osteogenesis. J Biol Chem. 2008; 283(11):7082–93. 20. Moniot B, Biau S, Faure S, Nielsen CM, Berta P, Roberts DJ, de Santa Barbara P. SOX9 specifies the pyloric sphincter epithelium through mesenchymal-epithelial signals. Development. 2004;131(15):3795–804. 21. Zehentner BK, Haussmann A, Burtscher H. The bone morphogenetic protein antagonist Noggin is regulated by Sox9 during endochondral differentiation. Dev Growth Differ. 2002;44(1):1–9. 22. Bergmann C, Senderek J, Anhuf D, Thiel CT, Ekici AB, Poblete-Gutierrez P, van Steensel M, Seelow D, Nu¨rnberg G, Schild HH, Nu¨rnberg P, Reis A, Frank J, Zerres K. Mutations in the gene encoding the Wnt-signaling component R-spondin 4 (RSPO4) cause autosomal recessive anonychia. Am J Hum Genet. 2006; 79(6):1105–9. 23. Blaydon DC, Ishii Y, O’Toole EA, Unsworth HC, Teh MT, Ru¨schendorf F, Sinclair C, Hopsu-Havu VK, Tidman N, Moss C, Watson R, de Berker D, Wajid M, Christiano AM, Kelsell DP. The gene encoding R-spondin 4 (RSPO4), a secreted protein implicated in Wnt signaling, is mutated in inherited anonychia. Nat Genet. 2006;38(11):1245–7. 24. Kim JS, Ryoo ZY, Chun JS. Cytokine-like 1 (Cytl1) regulates the chondrogenesis of mesenchymal cells. J Biol Chem. 2007;282(40): 29359–67. 25. Neame PJ, Tapp H, Grimm DR. The cartilage-derived, C-type lectin (CLECSF1): structure of the gene and chromosomal location. Biochim Biophys Acta. 1999;1446(3):193–202. 26. Geering K, Be´guin P, Garty H, Karlish S, Fu¨zesi M, Horisberger JD, Crambert G. FXYD proteins: new tissue- and isoform-specific regulators of Na,K-ATPase. Ann N Y Acad Sci. 2003;986:388–94.

31. Phan MN, Leddy HA, Votta BJ, Kumar S, Levy DS, Lipshutz DB, Lee SH, Liedtke W, Guilak F. Functional characterization of TRPV4 as an osmotically sensitive ion channel in porcine articular chondrocytes. Arthritis Rheum. 2009;60(10):3028–37. 32. Clark AL, Votta BJ, Kumar S, Liedtke W, Guilak F. Chondroprotective role of the osmotically sensitive ion channel transient receptor potential vanilloid 4: age- and sex-dependent progression of osteoarthritis in Trpv4-deficient mice. Arthritis Rheum. 2010;62(10): 2973–83. 33. Bjarnado´ttir TK, Fredriksson R, Ho¨glund PJ, Gloriam DE, Lagerstro¨m MC, Schio¨th HB. The human and mouse repertoire of the adhesion family of G-protein-coupled receptors. Genomics. 2004;84(1):23–33. 34. Borzı` RM, Mazzetti I, Marcu KB, Facchini A. Chemokines in cartilage degradation. Clin Orthop Relat Res. 2004; (427 Suppl):S53–61. 35. Huang da W, Sherman BT, Lempicki RA. Systematic and integrative analysis of large gene lists using DAVID bioinformatics resources. Nat Protoc. 2009;4(1):44–57. 36. Salminen HJ, Sa¨a¨ma¨nen AM, Vankemmelbeke MN, Auho PK, Pera¨la¨ MP, Vuorio EI. Differential expression patterns of matrix metalloproteinases and their inhibitors during development of osteoarthritis in a transgenic mouse model. Ann Rheum Dis. 2002; 61(7):591–7. 37. Han Y, Lefebvre V. L-Sox5 and Sox6 drive expression of the aggrecan gene in cartilage by securing binding of Sox9 to a far-upstream enhancer. Mol Cell Biol. 2008;28(16):4999–5013. 38. Zhu M, Tang D, Wu Q, Hao S, Chen M, Xie C, Rosier RN, O’Keefe RJ, Zuscik M, Chen D. Activation of beta-catenin signaling in articular chondrocytes leads to osteoarthritis-like phenotype in adult beta-catenin conditional activation mice. J Bone Miner Res. 2009; 24(1):12–21. 39. Yuasa T, Kondo N, Yasuhara R, Shimono K, Mackem S, Pacifici M, Iwamoto M, Enomoto-Iwamoto M. Transient activation of Wnt/{beta}catenin signaling induces abnormal growth plate closure and articular cartilage thickening in postnatal mice. Am J Pathol. 2009;175(5): 1993–2003. 40. Kondo N, Yuasa T, Shimono K, Tung W, Okabe T, Yasuhara R, Pacifici M, Zhang Y, Iwamoto M, Enomoto-Iwamoto M. Intervertebral disc development is regulated by Wnt/beta-catenin signaling. Spine (Phila Pa 1976). 2011;36(8):E513–8. 41. Annunen S, Paassilta P, Lohiniva J, Pera¨la¨ M, Pihlajamaa T, Karppinen J, Tervonen O, Kro¨ger H, La¨hde S, Vanharanta H, Ryha¨nen L, Go¨ring HH, Ott J, Prockop DJ, Ala-Kokko L. An allele of COL9A2 associated with intervertebral disc disease. Science. 1999;285(5426):409–12. 42. Jim JJ, Noponen-Hietala N, Cheung KM, Ott J, Karppinen J, Sahraravand A, Luk KD, Yip SP, Sham PC, Song YQ, Leong JC, Cheah KS, Ala-Kokko L, Chan D. The TRP2 allele of COL9A2 is an age-dependent risk factor for the development and severity of intervertebral disc degeneration. Spine (Phila Pa 1976). 2005; 30(24):2735–42.

27. Lucero HA, Kagan HM. Lysyl oxidase: an oxidative enzyme and effector of cell function. Cell Mol Life Sci. 2006;63(19–20):2304–16.

43. Higashino K, Matsui Y, Yagi S, Takata Y, Goto T, Sakai T, Katoh S, Yasui N. The alpha2 type IX collagen tryptophan polymorphism is associated with the severity of disc degeneration in younger patients with herniated nucleus pulposus of the lumbar spine. Int Orthop. 2007;31(1):107–11.

28. Stelzer C, Brimmer A, Hermanns P, Zabel B, Dietz UH. Expression profile of Papss2 (30 -phosphoadenosine 50 -phosphosulfate synthase 2) during cartilage formation and skeletal development in the mouse embryo. Dev Dyn. 2007;236(5):1313–8.

44. Seki S, Kawaguchi Y, Mori M, Mio F, Chiba K, Mikami Y, Tsunoda T, Kubo T, Toyama Y, Kimura T, Ikegawa S. Association study of COL9A2 with lumbar disc disease in the Japanese population. J Hum Genet. 2006;51(12):1063–7.

29. Coste B, Mathur J, Schmidt M, Earley TJ, Ranade S, Petrus MJ, Dubin AE, Patapoutian A. Piezo1 and Piezo2 are essential components of distinct mechanically activated cation channels. Science. 2010; 330(6000):55–60.

45. Kawaguchi Y, Osada R, Kanamori M, Ishihara H, Ohmori K, Matsui H, Kimura T. Association between an aggrecan gene polymorphism and lumbar disc degeneration. Spine (Phila Pa 1976). 1999;24(23):2456–60.

30. Boulter C, Mulroy S, Webb S, Fleming S, Brindle K, Sandford R. Cardiovascular, skeletal, and renal defects in mice with a targeted disruption of the Pkd1 gene. Proc Natl Acad Sci U S A. 2001;98(21): 12174–9.

2524

HENRY ET AL.

46. Solovieva S, Noponen N, Ma¨nnikko¨ M, Leino-Arjas P, Luoma K, Raininko R, Ala-Kokko L, Riihima¨ki H. Association between the aggrecan gene variable number of tandem repeats polymorphism and intervertebral disc degeneration. Spine (Phila Pa 1976). 2007; 32(16):1700–5.

Journal of Bone and Mineral Research

47. Cong L, Pang H, Xuan D, Tu GJ. Association between the expression of aggrecan and the distribution of aggrecan gene variable number of tandem repeats with symptomatic lumbar disc herniation in Chinese Han of Northern China. Spine (Phila Pa 1976). 2010;35(14):1371–6. 48. Mashayekhi F, Shafiee G, Kazemi M, Dolati P. Lumbar disk degeneration disease and aggrecan gene polymorphism in northern Iran. Biochem Genet. 2010;48(7–8):684–9. 49. Seki S, Kawaguchi Y, Chiba K, Mikami Y, Kizawa H, Oya T, Mio F, Mori M, Miyamoto Y, Masuda I, Tsunoda T, Kamata M, Kubo T, Toyama Y, Kimura T, Nakamura Y, Ikegawa S. A functional SNP in CILP, encoding cartilage intermediate layer protein, is associated with susceptibility to lumbar disc disease. Nat Genet. 2005;37(6):607–12. 50. Kagan HM, Trackman PC. Properties and function of lysyl oxidase. Am J Respir Cell Mol Biol. 1991;5(3):206–10. 51. Kurima K, Warman ML, Krishnan S, Domowicz M, Krueger RC Jr, Deyrup A, Schwartz NB. A member of a family of sulfate-activating enzymes causes murine brachymorphism. Proc Natl Acad Sci U S A. 1998;95(15):8681–5. 52. Faiyaz ul Haque M, King LM, Krakow D, Cantor RM, Rusiniak ME, Swank RT, Superti-Furga A, Haque S, Abbas H, Ahmad W, Ahmad M, Cohn DH. Mutations in orthologous genes in human spondyloepimetaphyseal dysplasia and the brachymorphic mouse. Nat Genet. 1998;20(2):157–62. 53. Nakamura Y, Yamamoto K, He X, Otsuki B, Kim Y, Murao H, Soeda T, Tsumaki N, Deng JM, Zhang Z, Behringer RR, Crombrugghe B, Postlethwait JH, Warman ML, Nakamura T, Akiyama H. Wwp2 is essential for palatogenesis mediated by the interaction between Sox9 and mediator subunit 25. Nat Commun. 2011;2:251. 54. Lyons G, Eisenstein SM, Sweet MB. Biochemical changes in intervertebral disc degeneration. Biochim Biophys Acta. 1981;673(4): 443–453. 55. Watanabe H, Nakata K, Kimata K, Nakanishi I, Yamada Y. Dwarfism and age-associated spinal degeneration of heterozygote cmd mice defective in aggrecan. Proc Natl Acad Sci U S A. 1997;94(13):6943–7. 56. Hou B, Kolpakova-Hart E, Fukai N, Wu K, Olsen BR. The polycystic kidney disease 1 (Pkd1) gene is required for the responses of osteochondroprogenitor cells to midpalatal suture expansion in mice. Bone. 2009;44(6):1121–33. 57. Harteneck C, Reiter B. TRP channels activated by extracellular hypoosmoticity in epithelia. Biochem Soc Trans. 2007;35(Pt 1):91–5.

Journal of Bone and Mineral Research

58. Guilak F, Leddy HA, Liedtke W. Transient receptor potential vanilloid 4: The sixth sense of the musculoskeletal system?. Ann N Y Acad Sci. 2010;1192:404–9. 59. Rock MJ, Prenen J, Funari VA, Funari TL, Merriman B, Nelson SF, Lachman RS, Wilcox WR, Reyno S, Quadrelli R, Vaglio A, Owsianik G, Janssens A, Voets T, Ikegawa S, Nagai T, Rimoin DL, Nilius B, Cohn DH. Gain-of-function mutations in TRPV4 cause autosomal dominant brachyolmia. Nat Genet. 2008;40(8): 999–1003. 60. Lamande´ SR, Yuan Y, Gresshoff IL, Rowley L, Belluoccio D, Kaluarachchi K, Little CB, Botzenhart E, Zerres K, Amor DJ, Cole WG, Savarirayan R, McIntyre P, Bateman JF. Mutations in TRPV4 cause an inherited arthropathy of hands and feet. Nat Genet. 2011;43(11):1142–6. 61. Soranzo N, Rivadeneira F, Chinappen-Horsley U, Malkina I, Richards JB, Hammond N, Stolk L, Nica A, Inouye M, Hofman A, Stephens J, Wheeler E, Arp P, Gwilliam R, Jhamai PM, Potter S, Chaney A, Ghori MJ, Ravindrarajah R, Ermakov S, Estrada K, Pols HA, Williams FM, McArdle WL, vanMeurs JB, Loos RJ, Dermitzakis ET, Ahmadi KR, Hart DJ, Ouwehand WH, Wareham NJ, Barroso I, Sandhu MS, Strachan DP, Livshits G, Spector TD, Uitterlinden AG, Deloukas P. Meta-analysis of genome-wide scans for human adult stature identifies novel Loci and associations with measures of skeletal frame size. PLoS Genet. 2009;5(4):e1000445. 62. Liu JZ, Medland SE, Wright MJ, Henders AK, Heath AC, Madden PA, Duncan A, Montgomery GW, Martin NG, McRae AF. Genome-wide association study of height and body mass index in Australian twin families. Twin Res Hum Genet. 2010;13(2):179–93. 63. Zhao J, Li M, Bradfield JP, Zhang H, Mentch FD, Wang K, Sleiman PM, Kim CE, Glessner JT, Hou C, Keating BJ, Thomas KA, Garris ML, Deliard S, Frackelton EC, Otieno FG, Chiavacci RM, Berkowitz RI, Hakonarson H, Grant SF. The role of height-associated loci identified in genome wide association studies in the determination of pediatric stature. BMC Med Genet. 2010;11:96. 64. Palmgren T, Gro¨nblad M, Virri J, Seitsalo S, Ruuskanen M, Karaharju E. Immunohistochemical demonstration of sensory and autonomic nerve terminals in herniated lumbar disc tissue. Spine. 1996; 21(11):1301–16. 65. Freemont AJ, Peacock TE, Goupille P, Hoyland JA, O’Brien J, Jayson MI. Nerve ingrowth into diseased intervertebral disc in chronic back pain. Lancet. 1997;350(9072):178–81.

THE POSTNATAL ROLE OF Sox9 IN CARTILAGE

2525